We democratize data science competitions, allowing startups to sponsor challenges and award prizes accessible to them. We give startups access to a global talent pool to help them solve their problems using machine learning.
Many startups do not have the talent or great resources to experiment with sophisticated data science or machine learning algorithms to find innovative solutions to their problems.
On our platform startups can sponsor a competition by giving an affordable prize money to a global talent pool that competes (for a period of time) to deliver the best solution to that problem.
The people competing for the prize are data scientists who give you an algorithm (machine learning model) to be used exclusively by your startup to solve the problem.
We have a platform capable of hosting competitions, uploading files, evaluating them and ranking the best results in the competition. You will also be able to know at all times the rankings of all competitors.
$ 499+20% Fee, Per competition Start Now
$ 999+30% Fee, Per competition Start Now
$ 0Contact Us Request Demo
Problem complexity based on our
problem framing tool*
|Low-level complexity||Mid-level complexity||High-level complexity|
Competition setup based on our
problem framing tool*
|Self-setup||Our team support you||Our team support you|
Cleaning and organizing the data sets
based on our problem framing tool*
|You must have the dataset ready||Our team support you||Our team support you|
|Prize Winners Prize distribution||Only to the first place||First two places||First three places|
|Time Competition length||4 weeks||8 weeks||3-6 months|
|Solution What you get||Top 5 competitors' algorithms||Top 20 competitors' algorithms||Top 30 competitors' algorithms|
|Our Fee DataSource.ai Fee||+20%||+30%||+30%|
|Access to the whole competitors Access to our entire talent pool||
|Intellectual Property Code Intellectual Property||
|Curriculum Vitae Access to competitors' CVs||
|API Integration Solution integration via API||
|Featured Featured space on the platform||
The level of complexity can be given by different and multiple factors, the most common are given by the type of algorithm that should be used to deliver a solution. It also depends on the nature of the data, the format in which it is found and the problem to be solved. For this we have a problem framing tool , which will tell you which is the right plan for you, and if you will need our help for the definition of the problem, order of the data and configuration of the competition
We have a a problem framing tool at your disposal, which can guide you and help you understand how we calculate the complexity of the competition. If you are not clear about this, we can help you with the whole process if you become a "Pro Sponsor".
The competition will need all the details necessary for the competitors to create their models and execute the task correctly. You will need to be very clear about the problem you want to solve, the algorithms the participants could use, but above all: to have the data ready to be launched in the competition. You must have the training data, test data and private data with which we will measure the effectiveness of the models sent by the participants. In addition, the evaluation metrics to be used must be clear, as well as other simple details such as the description of the competition. It is also very important to be clear about how you will implement the winning solution. If you are not clear about this, we can help you with the whole process if you become a "Pro Sponsor".
The data will be related to the problem you want to solve. You should already have it collected (internally or externally). You should be clear about whether it is data that is already labeled or not. It should also be clean, organized, transformed, and largely ready so that participants can focus on building a model around it. If you are not clear about this, we can help you with the whole process if you become a "Pro-Sponsor".
The process of collecting, cleaning, wrangling and transforming data is a process that is performed by the startup prior to the competition. You must have the talent/time/resources to do it. Our platform focuses on delivering the best machine learning model that our community can do in a given period of time. However, if you do not have the talent/time/resources to do it on your own, we can help you with the whole process if you become a "Pro Sponsor".
The winners are defined by our platform. The platform has an automatic evaluator, which is in charge of "ranking" the solutions sent by the competitors in real time. Competitors normally send a considerable amount of models for each competition. They do this in order to measure, iterate and find the best solution. However, at the end of the competition the first 20 competitors in the ranking must send us the complete original model so that we can objectively evaluate the solution, thus avoiding that the models are overfitted. With this, we make sure to give you the best possible model and also give the prize to the right winner.
We understand that there are machine learning problems that are "simple" to solve, that the startup needs them as soon as possible, that it is not necessary to extend the competition for months, and that this way we keep the motivation high of the current competitors. This also allows competitors to generate recurring income, keep the focus on the platform and the problem to be solved, all of this for the benefit of the competition sponsored by your startup. Also the Pro Sponsor competition lasts 8 weeks, because the complexity tends to be higher, so the data scientists will need more time to solve it. And in "Enterprise" there could be competitions open for 3 or 6 months. It all depends on the complexity of the problem and the size of the final prize.
At the end you will receive a machine learning model. This model is an algorithm that will allow you to solve the problem that was setup at the beginning of the competition. This algorithm has all the intellectual property rights transferred by the winner (or winners), due to the prize money you have given them. This is in the terms and conditions of service, therefore they have accepted this condition. So your startup will be the only one to use this solution. The algorithm will be delivered in a Jupyter Notebook and/or Python file with the necessary code for its internal implementation.
Once you receive the solution you should implement it in your internal systems, and maybe in real time. This is where you will find the usefulness of the solution delivered at the end of the competition. If you are not clear on how to do this, we can help you with the whole process if you become a "Pro Sponsor".
For the "Basic Sponsor" prize plan we charge a 20% fee which is added to the value
of the competition. This means that the winning competitor receives $499, DataSource.ai
receives $99.8, and the startup pays a total of $598.8. For the "Pro Sponsor" prize plan
we charge a 30% fee which is added to the value of the competition. This means that
the winners receive $999, DataSource.ai receives $299.7, and the startup pays a total
of $1,298.7. In this plan the commission is higher because the competition meets some
(or all) of these characteristics:
Yes, in the "Pro Sponsor" and "Enterprise" plans the startup has the right to receive the Curriculum Vitae of the first 20 participants of the ranking.
No, at the moment we do not have the option of direct contact between the startup and the participants. If you have any questions about the competition, please write to us. Also if you want to contact them directly for a job offer please write to us here: [email protected]
Our talent pool consists of more than 1,500 data science experts, ranging from mathematicians, statisticians, programmers to data scientists, with all kinds of backgrounds: from self-taught to PhDs. These are the people who will compete to solve your problem!
We help startups in all sectors and countries to improve their data analysis, prediction, classification and deep learning processes via data science competitions.